Overview
At the outset of our project, our primary focus was on conducting wet experiments and practical applications, with little emphasis on hardware design. However, as we progressed, we recognized that our existing equipment was insufficient for our needs. It failed to support the complexities of multi-process reactions, resulting in poorly connected reactions and an increase in impurities and precipitation.
Understanding that others might face similar challenges in multi-step experiments, we decided to take the initiative to design our own hardware. Our equipment specifically addresses common issues associated with multi-process reactions, such as incomplete reactions and impurity susceptibility, by implementing tailored countermeasures for each problem. This approach not only creates a more suitable environment for our cultures but also enhances the hardware's compatibility for large-scale industrial production, following consultations with industry professionals.
Through ongoing digital simulations and field research, we continuously refined and optimized our hardware design. During discussions with stakeholders from a battery company, we identified limitations in their existing battery sorting process. In response, we developed an intelligent sorting system based on color recognition to meet their needs. This system allows for angle rotation based on color detection, enhancing sorting efficiency.
Not only does iGEM focus on experimental projects, but also emphasizes the connection and mutual influence between projects, society, and stakeholders. Therefore, our hardware design incorporates both experimental requirements and stakeholder needs, ensuring that our solutions are not only effective but also widely applicable.
We believe that synthetic biology should be accessible to as many people as possible and that science should evolve with each improvement. To support this, we have made relevant documentation available on our website, enabling a broader audience to utilize and enhance our devices.
Intelligent sorting system based on color recognition
Design purpose
After background research and communication with companies, we found that existing battery sorting still relies on manual sorting, which is not only inefficient, but also requires workers to perform a single boring and dangerous task. Prolonged exposure to chemical waste liquid leaking from used batteries can cause irreversible damage to workers' health. This is not what we want to see.We want to improve workers' working environment. So we decided to research an intelligent sorting system to solve this problem.
At the same time, we noticed a huge difference in the color of the appearance of different types of batteries. Both the Society of Automotive Engineers and the Japan Battery Association recommend that manufacturers use a uniform labeling standard when designing the appearance of electric vehicle batteries. The latest algorithms in computer vision research have the ability to recognize objects and materials based on features such as size, shape, color and texture. Recycling may be facilitated if manufacturers add labels, QR codes, or other machine-readable features to key battery components and substructures. In order to promote uniformity in the appearance of batteries and to facilitate their sorting and recycling, we decided to start with the aspect of color and build an intelligent system based on color recognition.
Design ideas
Fig. 1 Overall deisgn idea
We plan to use the visual recognition module to classify objects and return different value, and the motor will perform different operations according to the different data returned to realize the sorting of objects of different colors. At the same time, a uniform speed transfer system is designed to deliver blocks to the camera for recognition. Based on the concept of environmental protection and recycling, for parts that need to be customized, we use low-pollution, degradable materials for 3D printing or use waste, such as plastic bottles, so as to minimize the harm to the environment.
Fig. 2 Screenshot of Simulink simulation of the rotating part at a constant speed
The input signal of the system is a step signal with a final value of 1.5, and the PID controller Kd = 0 of the forward path is used to introduce PI control to optimize the transient and steady-state characteristics of the system. In the feedback loop, the PID controller Kp = 0, KI = 0, Kd = 1 is used to feed back the speed signal of the motor and play a role in stabilizing the speed.
Fig. 3 Screenshot of Simulink modeling of the classification rotation system
When the motor is connected to the load, the steady-state error increases, and the classification operation cannot be accurately realized, so we add a PID controller to the forward path, and introduce the PI algorithm, which combines the rapid response of proportional adjustment with the characteristics of eliminating the steady-state error in the integration link, and realizes a better adjustment effect, so that the classification device can quickly and accurately make the corresponding rotation operation after receiving the signal, and can reset to the initial position.
Mechanical structure
We used 2 Qube-Servo2 motors, 1 camera and mount, and self-made components to build our overall mechanical structure. At the same time, we replace the sorted used batteries with small blocks of different colors.
Fig. 4 The overall physical structure of the sorting system
In this system, the camera is responsible for color recognition, and at the same time feeds back a value to the system for subsequent control; One of the two motors is responsible for the corresponding operation according to the different data fed back by the camera.
In addition, we also designed some hardware, including turntables, brackets, chassis, etc.
Fig. 5 1. Chassis 3D model drawing 2. 3D model of the bracket 3. Turntable 3D model drawing
There are 8 small holes dug in the upper turntable of this device. The block will be placed in the small hole.When the turntable rotates to the vicinity of the chassis hole with the block, the camera will identify the color of the block and output the corresponding signal to the classification rotation transpose, and the classification model makes the guide trough bracket rotate to the corresponding angle according to the return value .Then the block falls from the chassis into the corresponding box, so as to realize the classification function.
Visual model analysis
Fig. 6 Screenshot of the simulink simulation in the visual part
The software design of the vision module uses the image processing module provided by QUARC, which allows the vision module to recognize the target color in the image and determine the number of pixels in the image by adjusting the HSV parameters. With the output of the number of pixels, we can determine the color recognized by the vision module.
Fig. 7 Screenshot of the simulink simulation in the visual part
When the number of pixels of the signal is greater than the set threshold, the trigger can be triggered and the corresponding signal is output, the output falling edge when the number of red pixels is greater than the set threshold, the output rising edge when the number of yellow pixels is greater than the set threshold, and 0 under other conditions.
At the same time, we also found that the camera reacts differently to blocks of different colors, and it is less sensitive to yellow blocks, so we added a delay module to the back of the red block, so that the trough guide device has the same response speed for blocks of different colors.
Experimental results
Before building the actual control system, we first used Simulink in Matlab to test the motor simulation model, and the specific waveform diagram and description are as follows.
Fig. 8 1. Variation curve of motor position information with time in a constant rotation system 2. Variation curve of classified motor position information over time
According to the time-dependent curve of the position information output by the motor, it can be found that after the motor is started, a short acceleration process has passed, and then the slope of the curve is stable, which proves that the motor speed remains constant. Both the overshoot and steady-state errors of the system are approximately zero, which proves that the control system is stable.
According to the time-varying curve of the position information output by the motor, it can be found that after the step signal is input, the motor responds quickly, rotates a fixed angle, and keeps the current position unchanged. Both the overshoot and steady-state errors of the system are approximately zero, which proves that the control system is stable.
We set up the control system and tested it with small blocks of different colors. Through our counting, the system has achieved the sorting of 64 small blocks in 1 minute, of which 55 blocks are sorted correctly, with an accuracy rate of more than 90%.
Test video
Summary
The vision module based on the HSV color space and the motion module based on the PID algorithm and other control algorithms can accurately and stably divide the blocks into three categories of blocks: red, yellow and other colors, which can be used for enterprise battery sorting. In addition, this device can be used in other fields, such as agriculture, industry, etc., to replace labor-intensive and tedious work. We put the relevant documents at the end for future iGEM teams to learn.
Multi-process reaction system
Design purpose
During the experiment, we found that the existing hardware could not meet the needs of our experiment. At the same time, we also found that in the field of metal ion wastewater recycling, there are few experimental instruments that combine the two processes of reaction and adsorption. Therefore, we decided to design a comprehensive experimental hardware that combines "reaction-sedimentation-adsorption". This can be used not only for our experiments and the industrialization of our projects, but also for other similar fields.
Design ideas
Fig. 9 Overall design idea
We used the engineering software Solidworks to design the overall hardware equipment, and then we used Cosmol to simulate the adsorption process. After receiving the results, we consulted with experts in reactor design, who commented on our design, and we improved our design based on their recommendations, and continued to iterate.
Hardware structure
Fig. 10 Hardware perspective
Two small reactors on the left are used for the fermentation of Aspergillus niger to produce citric acid, and the other for the display of GOx on the surface of the yeast, thus using glucose to produce gluconic acid and hydrogen peroxide. In both units, we have added a stirring device to fully react. Considering that Aspergillus niger is aerobic, we added an oxygen-permeated tube to the stirring device, and there are holes in the leaves to allow the bacteria to fully contact oxygen. In order to avoid contamination, the yeast's fermentation tank is not oxygenated.
Then the liquid produced by the two devices is passed into the reaction device and reacted with the black powder, and we still add a stirring device to make the reaction fully carried out. We take into account the various impurities in the black powder that do not react with the acid, so we add a three-stage filter to obtain a clear night with different metal ions.
The liquid flows through four tanks in turn, each tank adsorbs different metal ions, and finally the remaining liquid flows into the reaction tank that reacts with the black powder at the beginning, reflecting our design concept of internal circulation.
Fig. 11 Hardware side view & schematic diagram
Fig. 12 Side view of the "Reaction + Sedimentation" device
Our filtration unit is based on the filter tank commonly used in sewage treatment, we design a three-stage filter tank with cavities on each layer, and when the liquid is filled with the primary filter tank, it will flow into the next stage, so as to fully settle the impurities in the liquid.
Fig. 13 Side view of adsorption device
For the design of the adsorption unit, we added a negative to each tank, which was wrapped with a non-woven fabric of yeast to absorb ions. Each negative is permeable to the liquid, ensuring that the liquid flows into the next tank.
Structure diagram
Fluid simulation
We simulated the adsorption process using Cosmol, and we added 16 negatives to the adsorption column with a radius of 0.29 dm and a height of 2 dm, each with a surface material that binds ions, and the final result is shown in the GIF.
Fig. 14 Animation of simulation results
At the same time, we also selected a point and a line in the model. The local concentration change was collected, and it can be seen that for a point in the adsorption column, the concentration change is a straight line that rises by one point, and finally reaches saturation; For a line, due to the placement of the engineered bacteria in the middle, the concentration is not a perfect straight line but rather a broken line.
Fig. 15 Local data result graph
Iterative process
Our research process was not all smooth sailing, and the existing model was improved many times. Our first assumption is as follows, and it can be seen that the device setup is relatively simple in terms of reaction and sedimentation, without considering how to make the reaction completely carried out and how to settle sufficiently, only adding a baffle to reduce the flow rate of the liquid.
Fig. 16 First design drawing
For this hardware design, we consulted Mr. Xu Xiyan, a member of the Environmental Protection Committee of the Chemical and Engineering Society of China, a senior member of the Chinese Nuclear Society and the Chinese Society for Environmental Sciences. As an expert in the field of wastewater treatment and reactor design,he pointed out that it was not enough for us to consider only the leaching and adsorption of the reaction waste liquid, we also need to consider the mineralization reaction waste liquid, so we decided to pour the remaining liquid of the mineralization reaction into the reaction tank in the middle to improve the adsorption rate of ions.
Fig. 17 Group photo with Mr. Xu Xiyan
In addition, we interviewed Prof. Bo Wang and Prof. Xiao Feng from the School of Chemistry and Chemical Engineering at the Beijing Institute of Technology, who suggested that we add a stirring device to make the reaction fully perform. At the same time, from the perspective of low-carbon environmental protection, they suggested that we use metal-organic frameworks (MOFs) for carbon dioxide absorption and reduce carbon emissions. We reflected on their suggestion and decided to add a stirring device, and at the same time, MMCF-2 was added to the yeast growth environment and the stirring blades of the Aspergillus niger reaction device to absorb the carbon dioxide emitted by the growth of the culture. At the same time, we note that carbon dioxide is required at this stage of mineralization, and the carbon dioxide absorbed by the MOF can be discharged in this reaction by appropriately increasing the temperature, so as to achieve the carbon cycle within the project.
Fig. 18 Prof. Wang Bo and Prof. Feng Xiao (side-by-side)
After improving our hardware design, we visited Mr. Xu Xiyan again. While acknowledging the improvements that have been made, he points out that the existing installation is not sufficient to achieve the desired results, and suggests that we refer to the settling tank for wastewater treatment. However, we noticed that the volume of the wastewater sedimentation tank was large and could not be directly copied, so we fused the wastewater sedimentation tank with our sedimentation tank. The new sedimentation tank has three layers, and the liquid can only flow into the next layer when it is filled with one layer, which makes a pool can achieve three levels of sedimentation, which greatly lengthens the sedimentation time and improves the sedimentation effect.
Fig. 19 Reference settling tank and Improved settling tank
Future improvement direction
After literature research, we found that the effect of using a double helix structure similar to DNA in the adsorption column may be better than using ordinary round negatives. We used Cosmol to perform digital simulations, and the results were in line with our expectations, and the time required for the adsorption column surface to reach saturation was significantly shorter than that of ordinary discs using the spiral-shaped negatives, and we plan to apply this finding to the real world in the future.
Fig. 20 Optimization of adsorption columns
Biomineralization Supporting Control System
Design Objective
When designing a multi-process reaction system, we did not initially include a device for biomineralization. As the hardware work neared completion, based on industry research, we began addressing the lack of hardware design for mineralization equipment. To link biomineralization with bio-adsorption, we designed a supporting control system to strengthen the connection between the two experiments and make the regulation process more intelligent.
Design Concept
We envisioned adding a fluorescent protein gene and a metal ion-repressible promoter to the Pichia pastoris used in the adsorption experiment. When the metal-binding peptides become saturated with ions and the concentration of metal ions in the solution decreases, the fluorescent protein gene would be expressed, causing the yeast to gradually turn red. At this point, the sensor detects the color change and reacts by moving the Pichia pastoris-coated films into the mineralization experiment solution using a servo motor.
We designed the system to recognize three colors: red, green, and blue. Upon detecting these colors, the system rotates 60 degrees, 120 degrees, and 180 degrees, respectively, and adjustments can be made based on wet lab experiment design. Part of the code is shown below.
Fig. 21 Part of the code
Overall Structure
We used a breadboard, a TCS34725 sensor, a servo motor, an STM32F103C8T6 microcontroller, and several DuPont wires.
The breadboard serves as the platform, carrying the DuPont wires and multiple electronic components.
The microcontroller we selected is the STM32F103C8T6. We preloaded the program onto the microcontroller, which processes input values and returns different outcomes.
Fig. 22 STM32F103C8T6
The servo motor is the only power component we used. It rotates to different angles based on the received commands.
Fig. 23 Servo Motor
The TCS34725 sensor is responsible for detecting different color signals. When it identifies red, green, or blue, it returns different values. These values are processed by the preloaded program, which sends different instructions to the servo motor, causing it to rotate.
The entire system is powered via a USB connection to a computer, eliminating the need for batteries.
Demonstration of Results
Demonstration video
Conclusion
While we previously designed an intelligent sorting system based on color recognition, the principle and purpose of this system are quite different. We are very pleased that the hardware we designed fits well with the project's experiments, which will benefit the future application of the project in experiments. Moreover, this system has great potential for improvement and could be applied in other fields in the future.
